SEFOP: a novel approach to data path module placement

Abstract

This paper is concerned with utilizing regularity information of data path modules in placement and shows outstanding results. Similarity Extraction is applied first on a data path module given a set of slice numbers, a column number, and IO pin ordering followed by a three-level Force-Oriented Placement for determining relative locations of cells in the form of a cell matrix. exact pin locations, cell flipping, and row/column spacing are then estimated based on the cell placement. Experiments on cell-based designs demonstrate its performance over various generate-purpose approaches at far less CPU time on data path module designs.